1: \begin{abstract}

2: \setlength{\parindent}{0cm}

3: 

4: The problem of matching unlabelled point sets using Bayesian inference is considered. 

5: Two recently proposed models for the likelihood are compared, based on the Procrustes

6: size-and-shape and the full configuration. 

7: Bayesian inference is carried out for matching point sets using 

8: Markov chain Monte Carlo simulation.  An improvement to the existing Procrustes algorithm is proposed which improves 

9: convergence rates, using occasional large jumps in the burn-in period. 

10: The Procrustes and configuration methods are compared in 

11: a simulation study and using real data, where it is of interest to estimate the 

12: strengths of matches between protein binding sites. The performance of both methods

13: is generally quite similar, and a connection between the two models is made

14: using a Laplace approximation.

15: \end{abstract}

16: